A ROBUST PERCEPTUAL HASH FUNCTION BASED ON DWT-SVD

Perceptual hash functions are commonly used in multimedia technologies. These functions are used in identification, image indexing, copying, moving object detection and forensic computing. These functions generally consist of three parts. These are preprocessing, feature extraction and generating hash function. In this article, a robust perceptual hash function is proposed with using Discrete wavelet transform (DWT) and Singular value decomposition (SVD) methods. During the preprocessing stage, the image is transformed from the color form to the gray form and gaussian filter is applied. Then, the central symmetric local binary pattern (CSLB) was applied to the image for reducing 8 bits to 4 bits. 3 level DWT is applied to this images and LL3 band was obtained. In this band, SVD was applied with using 4 × 4 non-overlapping blocks and the U, S and V matrices were obtained. The feature set is obtained using the largest value of the S matrix. The obtained feature set is cascaded to calculate the perceptual hash value of 256 bit length. Various attacks have been applied to calculate the performance of the proposed method. Experimental results show that proposed perceptual hash function is robustness.